undefined

Reporting a Kruskal-Wallis H Test in APA 7 Style

The Kruskal-Wallis H test is the non-parametric alternative to a one-way ANOVA, used when your dependent variable doesn't meet normality assumptions or is ordinal. Here's how to report it correctly in APA 7 format.

The basic format

H(df) = [statistic], p = [value], followed by an effect size (epsilon-squared or eta-squared-H), and post-hoc detail if the result is significant.

Example:

A Kruskal-Wallis H test showed a statistically significant difference in [DV] across the three groups, H(2) = 8.42, p = .015, ε² = .11. Dunn's post-hoc tests with Holm-Bonferroni correction indicated group A scored significantly higher than group C (p = .012), but no other pairwise differences were significant.

Three things people consistently get wrong

1. Naming the post-hoc correction. Don't just write "p < .05" — state the correction method explicitly. Bonferroni is conservative; Holm-Bonferroni is generally preferred and worth naming.
2. Skipping the effect size. H and p alone aren't sufficient for APA 7 — report epsilon-squared (ε²) or eta-squared-H alongside them.
3. Miscounting df. df = k − 1 (number of groups minus 1), not sample size minus 1.

Why this matters beyond style compliance

Reviewers and thesis committees increasingly check for correction-method transparency and effect sizes as a proxy for statistical rigor — omitting them is a common reason for "minor revisions" requests.

---

Disclosure: I built statmate.org, a tool that runs the test and generates this exact APA-formatted writeup automatically. Sharing the manual reasoning here regardless, since knowing the rule matters more than the tool.